can
First, the sample period is about 5.01 seconds, so there are 600/5 = 120 samples per 10 minutes. To maintain best accuracy, don't do anything with the PCE800 cumulative energy column for the input - it is integrated most accurately inside the meter. Next you want to construct the per-sample energy in the output based on the average temperature difference between 2 samples, the average flow between two samples, the time between samples (subtract the time stamps), and the 4.184 J/g/C. Then accumulate the per-sample energy to get the cumulative energy output curve. Now you have the cumulative input energy (provided by the PCE800) and the calculated output energy curve. For each 10 minutes of time (120 samples) centered on a sample, compute the energy change during that period in the output and in the input. Then the 10 minute rolling average COP is Delta-E_out/Delta-E_In.
Also note that when the Omega flow meter had no flow for about the first 21 samples, it registered an output of approximately 0.0212. This appeared to be an offset in the measurement, so I subtracted this from all samples so that with no flow, the signal would read 0.000. The water should actually have been flowed through the heat exchanger for a while until the inlet and the outlet temperatures were the same before turning on the reactor.
Your curve doesn't look quite the same, probably mostly from the different number of samples in the average, and from not correcting the flow meter signal (which I didn't mention before).